American Political Science Review Vol. 98, No.

3 August 2004

Presidentialism and Accountability for the Economy

in Comparative Perspective
DAVID SAMUELS University of Minnesota

T
o what extent do the institutions of presidentialism allow voters to hold governments accountable?
Powell and Whitten (1993) suggested that voter capacity to sanction is strong when clarity of
government responsibility for outcomes is clear, and vice versa. I argue that clarity of responsi-
bility functions differently under presidentialism and that presidentialism generates particular forms of
accountability. In general, electoral sanctioning is weak in nonconcurrent elections, which do not occur
under parliamentarism, but is stronger in concurrent elections. In concurrent executive elections the
clarity of responsibility does not attenuate the economys impact on the vote. Yet in concurrent legislative
elections both partisan and institutional variables diffuse responsibility for economic performance. Thus
under many common institutional and partisan formats, voters sanction presidents to a greater degree
than legislators for the same phenomenon. These findings elucidate the conditions under which we
might observe accountability similar to what we find in some parliamentary systems or a more uniquely
presidentialist dual democratic legitimacies of the kind Linz (1994) imagined.

he degree to which voters can hold elected offi- ability by enhancing identifiability and allowing vot-

T cials to accounts remains a central concern of

political science. Most research has explored
established democracies, which all happen to be par-
ers to hold presidents and legislators accountable for
different things, which may be normatively a good
thing (Perrson, Roland, and Tabellini 1997; Samuels
liamentary systems, with the exception of the United and Shugart 2003; Shugart and Carey 1992).
States. Democracys spread around the world has I seek to discover the ways and extent to which
increased the number of presidential systems, which we observe accountability for the economy under
generates an important new set of questions about presidentialism. Hundreds of books and articles have
the relationship between economics and elections, and explored the relationship between economics and elec-
thus about well democracy can work: To what extent tions,2 and scholars have begun to reveal the ways in
do the institutions of presidentialism allow voters to which political institutions and party-system configu-
hold the government accountable? To what extent do rations mediate this connection.3 Yet how accountabil-
we see party or government accountability in sepa- ity might work within a presidential system remains
ration of powers systems? That is, do voters reward largely unexplored, both empirically and theoretically.4
or punish incumbent presidents and legislators, and to Can presidential systems can live up to their demo-
the same degree? Do particular partisan dynamics or cratic promise, or does something about the sep-
institutional formats of presidentialism dim or enhance aration of powers inhibit accountability? Exploring
the prospects for accountability? this question could provide substantial insight into
Scholars have yet to provide answers to questions ongoing debates about the advantages or disadvan-
such as these, although substantial debate persists tages of different institutional formats for democratic
about the presidentialisms relative merits. Some schol- performance.
ars disdain presidentialism and suggest that the sepa-
ration of powers hinders accountability. For example, ACCOUNTABILITY AND CLARITY
Linz (1990, 1994) argued that presidentialisms dual OF RESPONSIBILITY
democratic legitimacies confuses voters and that pres-
idential autonomy inhibits party government ac- I define accountability as the electorates capacity
countability.1 Others question Linzs claim and suggest to reward or sanction incumbent politicians (Manin,
that presidentialism provides advantages for account- Przeworski, and Stokes 1999, 40).5 This paper thus

David Samuels is Associate Professor, Department of Politi-

2 For a recent review of this literature see Lewis-Beck and Stegmeier
cal Science, University of Minnesota, Minneapolis, MN 55455
([email protected]). 2000.
3 See, e.g., Anderson (1995, 2000) and Powell and Whitten (1993).
For assistance and comments on previous versions I thank
Octavio Amorim Neto, Teri Caraway, John Carey, Bill Clark, John 4 Cheibub and Przeworski (1999) provide a partial exception, al-
Freeman, Timothy Hellwig, Gabriel Negretto, Scott Morgenstern, though they define accountability differently, focusing on incum-
Shaheen Mozaffar, Alejandro Poire, Wendy Schiller, Phil Shively, bents survival in office.
and the participants at presentations at the University of Michigan, 5 I do not include the second clause of Manin, Przeworski, and
Fundacao Getulio Vargas Rio de Janeiro, New York University, Stokess definition, . . . so that those incumbents who act in the best
University of California at Irvine, Brown University, Instituto Tec- interest of citizens win reelection and those who do not lose them.
nologico Autonomo de Mexico, Centro de Investigaciones y Docen- Scholars typically find that economics determines about one-third of
cias Economicas in Mexico City, University of Minnesota, and the the vote (Dorussen and Palmer 2002, 4). Thus asking whether eco-
2003 APSA meetings in Philadelphia. nomic swings are sufficient to kick the bums out is asking too much.
1 See, e.g., Lijphart 1992, Manin, Przeworski, and Stokes 1999, and We expect a correlation between the economy and the vote, but a
ODonnell 1994. correlation between the economy and the removal of incumbents

1
Presidentialism and Accountability for the Economy August 2004

follows what Susan Stokes calls the normal economic be. This suggests that (to some degree at the least) they
voting research program. In this approach, voters ought to attempt to convince voters that they share
use the past performance of the government to predict a personal and political affinity with their successor-
future performance and see the government as respon- candidate.
sible for that performance (Stokes 2001, 13). Schol- Voters can reward a successful presidents party by
ars believe that voters capacity to reward or sanction electing his or her partys successor, or they can pun-
elected officials declines when they cannot discern re- ish the incumbents party by electing a rival partys
sponsibility for government performance. Powell and candidate. There is therefore no reason why voters
Whitten (1993) thus argued that voters are more likely could not hold political parties as a whole and pres-
to punish or reward incumbents for the state of the idential governments as a wholeeven those popu-
economy when clarity of responsibility for outcomes lated by term-limited politiciansaccountable in a ret-
is high. Given the separation of powers under presiden- rospective voting fashion. Thus while many problems
tialism, a pertinent question is therefore whether clar- inhere in the accountability relationship between cit-
ity of responsibility affects voters capacity to reward izens and governments, these problems are not lim-
or sanction presidential governments, in both executive ited to presidential systems, nor should we simply
and legislative elections. assume that presidential governments are necessarily
The theory of retrospective voting (Fiorina 1981; worse than other forms of government in terms of
Lewis-Beck 1988) suggests that electoral accountabil- accountability.
ity occurs because voters retrospectively judge whether Aggregate economic voting studies typically assume
governments have acted in their best interests and then that some voters consider national economic condi-
reward or sanction them appropriately. Yet several fac- tions before the election as part of their vote decision.
tors might hinder voters ability to sanction elected Voters attribute reward or blame to the incumbent gov-
officials. First, voters might not be able to punish in- ernment for those conditions, and as a result the vote
cumbents if no viable alternative exists (Anderson shifts. Yet electoral accountability declines when voters
2000; Manin et al. 1999). This problem is not limited cannot discern responsibility for government policy,
to presidentialism, and in the analysis I include only and voters are more likely to punish or reward the
competitive elections. Second, incumbent politicians incumbent government for the state of the economy
might strategically conceal information about policy if clarity of responsibility is high, and not if clarity of
responsibility. This is also a problem everywhere, and responsibility is low (Powell and Whitten 1993).
we can reasonably assume that voters in presidential Two related reasons explain this dynamic. The first
systems can potentially sanction governments, at least highlights incumbents strategic actions. All incum-
as much as they can anywhere else. In any case, it is bent governments tend to lose votes (Nannestad and
unreasonable to assume that incumbents facing com- Paldam 2002). Powell and Whitten surmised that in-
petitive elections conceal all information; instead, we cumbents know this and thus prefer to diffuse politi-
should test to see whether partisan and/or institutional cal responsibility, for example, by blaming those with
factors mediate the connection between economics and whom they might have shared control of government.
elections. A countrys institutional and/or partisan makeup can
Third, citizens will be able to sanction governments help politicians diffuse responsibility, or it can com-
only if politicians desire (and are eligible for) reelec- plicate such activity. The second factor emphasizes
tion or care about who succeeds them to office. Manin, voters ability or willingness to gather information:
Przeworski, and Stokes (1999, 48) suggest that limits on As institutional or partisan complexity increases, vot-
presidential reelection thus restrict voters capacity to ers will have greater difficulty discerning actual re-
hold governments accountable. Limits on reelection do sponsibility for government output, independently of
impede voters from rewarding or punishing a particular whether politicians attempt to diffuse responsibility or
individual incumbent president (or legislator), but they not. Regardless of which factor weighs more heav-
do not impede voters from sanctioning the incumbents ily, lower clarity of responsibility should insulate in-
party. In presidential systems, the vote for or against the cumbents from the factors that cause them to lose
party of a term-limited president or legislator is nec- votes.
essarily a judgment of both the individual incumbent
and the incumbents party. Manin and coworkers sug-
gestion assumes that incumbent presidents do not care
PRESIDENTIALISM AND CLARITY
about their legacy, have no interest in the continuity OF RESPONSIBILITY
of their policies, and do not care about their partys Perhaps clarity of responsibility and thus voters capac-
success once they have left office. Yet last-term presi- ity for sanctioning in presidential systems depends on
dents care about their place in history and, thus, care factors that scholars have identified for parliamentary
about their performance in office. Moreover, since their systems.6 Or presidentialism could affect how voters
legacy is profoundly shaped by who succeeds them to hold governments accountable in particular ways. For
office, they also care about who that person turns out to
6 Powell and Whitten (1993) argued that incumbent party or coali-
may not appear. Still, this does not mean that incumbents are not tion cohesion and the type of committee system are important for
responsive to voter sanction and do not try to change their policies clarity of responsibility. Information on those variables is unavailable
to avoid additional future sanctions. for most of the countries in this study.

2
American Political Science Review Vol. 98, No. 3

example, Manin, Przeworski, and Stokes (1999, 47) im- nonconcurrence should attenuate the impact of na-
ply that presidentialism obscures government respon- tional factors on the vote because (all else equal)
sibility generally and that coalition and divided gov- it focuses voters and candidates energies on local
ernments under presidentialism are particularly bad factors and candidates personal qualities, including
for accountability. In contrast, Persson, Roland, and but not limited to any potential incumbency advan-
Tabellini (1997) suggest that the separation of powers tages due to constituency service. In nonconcurrent
institutionalizes conflicts of interest between branches legislative elections we also tend to observe relatively
of government and thus encourages relatively greater less national party organization involvement and rel-
information revelation than parliamentary systems, atively more influence of local or regional political
which should aid voters efforts to discern policy re- messages.
sponsibility. A similar logic should hold for executive elections.
A few single-country studies support the notion that Nonconcurrence frees presidential candidates to cam-
differences exist in terms of voters ability and/or will- paign on entirely individualistic bases, meaning that the
ingness to sanction members of the same party who oc- election result will depend relatively more on voters
cupy different branches of government, such as Lewis- perceptions of the candidates personal qualities and
Beck (1997) on France and Erickson (1989, 1990) on relatively less on their partisan connections. That is,
the United States, but no cross-national research has we have reason to believe It might not be the econ-
yet confirmed or refuted these findings.7 How do voters omy, stupid in nonconcurrent presidential elections,
in presidential systems think about blame and reward? because national policy factors ought to matter rela-
Do they focus exclusively and always on the president, tively less.
regardless of who controls the legislature? Are leg- Another institutional factor that might mediate clar-
islators in some partisan or institutional environments ity of responsibility, in legislative elections, is the elec-
more or less likely to feel the brunt of the voters wrath? toral rule. Under presidentialism voters cast separate
Several factors might sharpen or blunt clarity of ballots for legislative and executive elections. Unlike
responsibility and thus voters propensity to reward in parliamentary systems, this offers the opportunity
or punish incumbents in presidential systems. The for substantially different dynamics to drive executive
most important factor is unique to presidentialism and legislative elections. The primary factor is whether
and is institutional: the electoral cycle, or the concur- the electoral system encourages or discourages local-
rence or nonconcurrence of elections. Others (see es- ism and/or individualism, giving legislators the choice
pecially Shugart 1995) have noted that the electoral whether to run with or against their president
cycle plays a key role in executivelegislative rela- and/or party (Carey and Shugart 1995; Shugart 1995).
tions. Here I suggest that it also mediates the ac- When the electoral rule for legislative elections encour-
countability relationship between voters and elected ages localism, the state of the economy should matter
officials. relatively less, all else equal. In contrast, when the rules
Concurrent elections under presidentialism most re- nationalize legislative campaigns, the economy should
semble elections under parliamentarism. When exec- matter more clearly and consistently.
utive and legislative elections are held simultaneously, Two other variables that we also find in parliamen-
candidates for executive and legislative elections have tary systems merit discussion. The first is coalition gov-
strong incentives to coordinate election campaigns, in ernment. Coalitions tend to obscure any one partys
order to benefit from campaign economies of scale, responsibility for government output (Anderson 1995,
coattail effects, and organizational advantages. That is, 2000). As in parliamentary systems, governing in coali-
concurrent elections tend to encourage a team effort, tion might insulate incumbent presidents or legislative
whereby politicians employ their national party orga- parties from the factors that might otherwise affect
nizations and national partisan messages to get out the vote swings for or against them. A contrast of vote
vote. This associates copartisans from both branches swings lends credence to this notion: Presidents who
of government in voters minds and encourages voters governed in coalition were relatively more insulated
to hold them accountable for the same things. Conse- from losses, losing on average 4.68% (N = 21), com-
quently, in concurrent elections voters have relatively pared to average losses of 6.31% (N = 54) for pres-
greater incentives to hold both branches coresponsible idents who did not govern in a coalition. A similar
for the economy, all else equal. result obtains for the legislative elections: Parties in
In contrast, when either executive or legislative coalition lost an average of 0.92% (N = 32), while
elections are held nonconcurrently, national policy is- parties that governed alone lost an average of 3.59%
sues such as the economy should matter relatively (N = 71).
less. For legislative elections, following the literature Second, minority government might insulate pres-
on U.S. congressional elections (e.g., Erickson 1990), idents and legislative parties from vote swings due to
economic performance. In unicameral presidential sys-
tems, minority government occurs when the presidents
7 Remmer (1991, 2001) is the only scholar who has explored the links
party or coalition does not control more than 50% of
between economics and elections in a cross section of presidential the seats in the single legislative chamber. In bicameral
systems. However, Remmer was not interested in the impact of po-
litical institutions and explored economic voting in Latin America
systems, minority government occurs when the pres-
generally. Moreover, she did not explore the impact of economics on idents party or coalition does not control more than
legislative elections. 50% of the seats in both chambers. When parties divide

3
Presidentialism and Accountability for the Economy August 2004

control over the branches of government, the incum- growth in year t 1 if the election was held in the
bent president can claim that all parties must share first six months of the year, and the change in GDP
responsibility for poor performance. Several studies growth in year t if the election was held in the last
confirm that minority parliamentary governments suf- six months of the year.11 I call this variable Economy.
fer smaller losses than majority governments (e.g., If the economy matters generally in presidential sys-
Powell and Whitten 1993 and Strm 1990). As with tems, this variable should have a positive and significant
coalition governments, minority presidential govern- coefficient.
ments also appear to insulate incumbents from losses. If clarity of responsibility diffuses or accentuates
Presidential parties running a candidate following a economic voting effects, we should observe variables
minority presidency lost on average 5.00% (N = 51), in multiple regression analysis that substitute for, de-
compared to average losses of 7.68% (N = 24) for crease, or wipe out the impact of the economy on vote
candidates following a majority president. However, it swings. The logic of the argument suggests that the ef-
remains to be seen whether the economy drives these fect of economics on elections is interactive with clarity
differences. of responsibility variables (Powell and Whitten 1993).
The factors explored thus far permit us to place Thus if the economy matters in general, we expect a
presidential systems on a continuum. At one end elec- positive and significant coefficient on Economy (all
tions are nationalized and always concurrent, and at else equal), and if another variable blurs the impact
the other end legislative elections are localized and of economic fluctuations on the vote, we would expect
often not concurrent with the executive election. In a negative and significant coefficient on the interacted
the real world, Costa Rica might fall on one end of term Economy OtherVariable when that variable is
the continuum while Colombia fell at the other end, added to the equation. The impact of the economy
and the United States would fall somewhere in the on the vote would then be the sum of the coefficient
middle. The question remains whether such variables on Economy plus the coefficient on the interacted
affect clarity of responsibility for incumbents in either variable.
branch of government. As per the discussion in the previous section, I ex-
plore the impact of the following independent vari-
ables: Concurrence, Electoral Rules (for legislative
CLARITY OF RESPONSIBILITY AND elections only), Coalitions, and Minority President, all
ECONOMIC VOTING: ANALYSIS operationalized as dummy variables. I enter each vari-
able directly and in an interacted term (with Economy).
I test the impact of clarity of responsibility and eco- The dependent variable in all regressions is the abso-
nomic conditions on elections in 23 presidential democ- lute percentage change in vote for the incumbent pres-
racies, defined as those in which both branches of idents party from the previous election. In the models
government are directly elected for fixed terms and for executive elections, the dependent variable is the
in which the head of government is not accountable vote change in the executive election, and in the models
to the legislature (Siaroff 2003).8 I employ aggregate for legislative elections, the dependent variable is the
economic and electoral data. I included a case if the vote change in the legislative election. In all models, I
country had had at least two consecutive democratic employ ordinary least squares regression with robust
electionsand if political and economic data were standard errors.12
available. This resulted in a sample of 75 executive In all regressions I also included two control vari-
elections and 103 legislative election.9 ables: the incumbent partys proportion of the vote
Scholars of economic voting typically employ one of in the previous election (Previous Vote), which should
the following three variables as the indicator of eco- have a negative impact because parties tend to lose
nomic performance: GDP growth, inflation, or unem- votes after being in office; and a dummy variable
ployment. I employ the percentage change in real per indicating whether or not the incumbent president
capita GDP growth (World Bank 2003).10 Following is running for reelection (Reelection). This variable
Pacek and Radcliff (1995), I use the change in GDP should return a positive coefficient because incumbents
running for reelection have advantages in terms of
8 I include cases from Siaroffs Category 2 as well as Category 6,
name recognition and organization relative to other
which functions essentially like Category 2. As Elgie (1999, 11)
notes, regime classification varies from one writer to another, but
the only case on which I differ from Siaroff is Bolivia, which Siaroff 11 Quarterly figures are unavailable for most countries in the sam-
classifies together with Switzerland. Despite a quirk in Bolivias ple. Other operationalizations could provide more nuance, but data
constitution, it functions as a pure presidential system. See Inter- limitations restrict my ability to conduct such tests. For example,
American Development Bank 2002; personal communication with because the time series is relatively short for most of these countries,
John Carey and Octavio Amorim Neto. I cannot explore the impact of shocks relative to long-term trends
9 Appendix 1 contains a list of the countries and elections. I explore (cf. Palmer and Whitten 1999).
differences across regime types, incorporating pure and semipresi- 12 I do not employ models with country dummies because of the
dential systems as well as parliamentary systems, in other work. numerous cross sections relative to the overall sample size in the
10 Unemployment information is unavailable for most of the coun- data set (with 23 cross sections I have an average of three executive
tries in the sample. As for inflation, the expectations about the impact elections and four legislative elections per country). I ran a series of
of inflation are far less clear than for that of GDP growth, supporting regressions with country dummies, and although some of the dum-
the use of the latter over the former, because it is unclear whether mies were significant, their inclusion did not substantially alter the
voters respond to the level or direction of inflation. results.

4
American Political Science Review Vol. 98, No. 3

Table 1 reveals that voters only sanction presidents

TABLE 1. Executive Elections, Baseline and their parties candidate when executive and legisla-
Models tive elections are concurrent. I suggested above that in
Independent Variable Concurrent Nonconcurrent nonconcurrent elections voters may focus on the candi-
Previous Vote .477 .438 dates as individuals and not as members of a party run-
(.177) (.580)
ning candidates for a set of offices. As a result they may
Economy 1.172 .517
(.311) (1.711) not connect either the performance of the incumbent
Reelection 8.680 president or the performance of the economy to their
(2.762) vote. For those who believe that executive elections
Constant 12.906 14.767 should flow from the state of the national economy, the
(8.682) (30.841) results for nonconcurrent elections are, at a minimum,
R2 .357 .123 discouraging, even if they are not conclusive.
N 61 14 A key question remains about the potential im-
Prob >F .000 .707 pact of incumbent presidents who are running for
Note: p = .01, p = .001 (one-tailed test). reelection. Cheibub and Przeworski (1999) and Manin,
Przeworski, and Stokes suggest that voters cannot en-
gage in retrospective economic voting when they liter-
ally cannot hold the person who has been responsible
candidates. Consequently, they (or their party, in the for managing the economy over the past few years (in
case of legislative elections, due to potential coattail theory at least) to accounts. Does incumbency enhance
effects) ought to receive a higher vote percentage, all accountability for the economy, and does the absence
else equal. of an incumbent running for reelection weaken ac-
Given my hypothesis about the importance of the countability?
electoral cycle, I first separate out concurrent from In 11 of the 61 concurrent elections an incumbent
nonconcurrent elections.13 Table 1 provides results for president ran for reelection. (No incumbents ran for re-
a comparison of the effects of the economy on the election in nonconcurrent executive elections.) Given
vote in both concurrent and nonconcurrent executive the results for concurrent elections in Table 1, we know
elections. that incumbency does matter on its own. In Table 2,
The first model tests for the impact of the Economy, I interact the Economy and Reelection variables to
Previous Vote, and Reelection in concurrent executive discover whether incumbency has both an independent
elections. The results conform to my expectations: The and a joint effect. If incumbency drives the accountabil-
vote in the current period is correlated with the vote in ity relationship, this variable should return a positive
the previous period, presidents running for reelection and significant coefficient. In addition, the Economy
can expect a higher vote share than candidates running variable should lose significance, if the hypothesis that
for their first term, and the state of the economy is the absence of an incumbent running for reelection
strongly correlated with the vote for the incumbent attenuates accountability is correct.
party. These results are robust to the exclusion of any The results do not support the extension of Cheibub
particular case. and Przeworskis and Manin and coworkers argument.
The results for nonconcurrent presidential elections Incumbency does not make for a tighter accountability
also conform to my expectations. In contrast to the relationship, at least for the economy. Voters hold in-
results for concurrent elections, neither the previous cumbent parties to accounts, even in executive elections
vote nor the state of the economy affects the vote swing and even if the incumbent is not running for reelection,
for the incumbent partys candidate. This result may
sound foreign to scholars and residents of the United
States, where presidential elections are always concur-
rent with legislative elections, and I acknowledge that TABLE 2. Executive Elections,
the results rest on relatively few cases. However, they Testing for Impact of Incumbents
are fairly robust: Note that the standard error on Econ- Independent Variable Concurrent
omy is three times the size of the coefficient. Future Previous Vote .476
work should return to this issue as more data become (.180)
available. Economy 1.156
(.333)
Reelection 8.125
(2.968)
13 Analysis of a correlation matrix for executive elections revealed Reelection Economy .266
that the Economy Concurrence variable was correlated at .91 with (.848)
the Economy variable (no other variables were correlated at higher Constant 12.870
than .79). Correlation at this level is problematic and makes the (8.803)
inclusion of both variables in one model unhelpful. I also attempted
to test for the impact of concurrence using a Proximity variable, R2 .358
which measures the proportion of the presidential (legislative) term N 61
that has elapsed relative to the legislative (presidential) term. Yet this Prob >F .000
variable, when interacted with the Economy variable, was similarly Note: p = .01, p = .001 (one-tailed test).
highly correlated with the Economy variable, at .92.

5
Presidentialism and Accountability for the Economy August 2004

tion all continue to be strongly correlated with the

TABLE 3. Concurrent Executive Elections vote, the relationship between the economy and the
Independent Variable Model 1 Model 2 vote does not change for incumbent-party presidential
Previous Vote .545 .555 candidates whose parties do not control a legislative
(.194) (.184) majority.
Economy 1.277 1.088
(.251) (.350)
Model 2 tests for any change when the president
Reelection 9.577 8.753 governs in coalition. This hypothesis tests whether vot-
(3.080) (3.003) ers distinguish situations when the president agrees
Minority President 3.555 to share power from situations where the president
(2.687) attempts to govern without other parties assistance,
Min Pres Economy .168 regardless of whether that coalition is majority or mi-
(.556) nority. This situation arose in 15 of the 61 concurrent
Coalition 3.601 cases. The null cases here include both single-party mi-
(3.844) nority and single-party majority presidents. As with
Coal Economy .227 Minority President, the introduction of this variable
(.808)
Constant 18.204 17.382
does not alter the results.
(10.149) (9.310) The results in Tables 1 through 3 reveal that in ex-
ecutive elections the only variable affecting the clarity
R2 .373 .367
of responsibility and thus the relationship between the
N 61 61
Prob >F .000 .000 economy and the vote is whether the presidential and
Note: p = .05, p = .01, p = .001 (one-tailed test). legislative elections are held concurrently. When this
occurs, the state of the economy strongly shapes pres-
idential election results: In each model, an increase
of approximately 1% in per capita GDP prior to the
as long as elections are concurrent. Linking partisan election results in an increase of a bit more than 1%
ties across branches of government, rather than the in the incumbent partys vote. The introduction of par-
presence or absence of an individual candidate, is rel- tisan clarity of responsibility variables does not affect
atively more important for overall accountability in this relationshipthat is, voter reward or sanction is
presidential systems. not shaped by the presence of minority government or
The next question is whether these results change coalition government.
when we add any of the other clarity of responsibility It is perhaps paradoxical that nonconcurrent ex-
variables. Table 3 presents several models for concur- ecutive elections do not clearly enhance clarity of
rent executive elections, adding in each variable and its responsibility for the economy. Nonconcurrent pres-
interaction term.14 For nonconcurrent elections, I have idential elections certainly concentrate attention on
too few cases to conduct similar and reliable analysis the presidential campaign. Yet while we might sup-
the results would be dependent on the influence of pose that nonconcurrent elections would concentrate
the existence of one case or another of (for example) voters goodwill or ire about the economy onto the
minority coalition government. presidential race, the results here suggest that solo
Model 1 tests for any change when the presidents presidential elections tend to eliminate the connection
party does not control a majority of the seats in the between the state of the economy and the vote. The
legislature, independently of whether the president findings here, although based on relatively few cases,
ultimately obtains a majority through coalition gov- tend to qualify the notion that voters always hold
ernment. This specifically tests whether voters dis- presidents to accounts for the state of the economy:
tinguish situations when the presidents team does Its the economy, stupid only holds clearly when
not unilaterally comprise a legislative majority rela- executive and legislative elections are held concur-
tive to all other situations. This situation arose in 41 rently. Nonconcurrent presidential elections may turn
of the 61 concurrent cases. The null cases are thus on something elsecandidates personal characteris-
those in which the presidents party controls more tics or other policy outputs. This issue merits further
than 50% of the seats. This hypothesis suggests that exploration.15
shared responsibility for government output might I now turn to legislative elections. To parallel the
dampen voters propensity to hold the incumbent pres- analysis for executive elections, Table 4 displays the
idents partys candidate responsible for the state of
the economy. Thus, the coefficient on the interacted
variable ought to be negative, offsetting the positive 15 With the addition of more cases and/or countries, I could test
expected coefficient on Economy. The results indicate for differences both across and within groups of countries (i.e.,
that although Previous Vote, Economy, and Reelec- countries that only have concurrent elections, countries that have
both concurrent and nonconcurrent elections, and countries that
have only nonconcurrent elections), to further explore the impact
of concurrence and incumbency. However, in this sample, only four
14 In this and the subsequent series of regressions, I could not run countries have executive elections that are sometimes concurrent
fully specified models because adding in the interacted variables and sometimes nonconcurrent (N = 12 total, 6 nonconcurrent), and
results in problematic degrees of multicollinearity (among the inter- only two countries presidential elections are always nonconcurrent
acted variables). (N = 8).

6
American Political Science Review Vol. 98, No. 3

TABLE 4. Legislative Elections: Baseline TABLE 5. Concurrent Legislative Elections

Models Independent
Independent Variable Concurrent Nonconcurrent Variable Model 1 Model 2 Model 3
Previous Vote .293 .186 Previous Vote .285 .346 .295
(.075) (.095) (.091) (.081) (.087)
Economy .799 .382 Economy 1.211 .952 .778
(.197) (.239) (.199) (.189) (.258)
Reelection 5.203 Reelection 7.772 4.881 4.913
(3.514) (3.259) (3.789) (3.609)
Constant 5.532 5.186 Minority President .046
(3.124) (4.754) (2.265)
R2 .321 .177 Min Pres .562
N 62 41 Economy (.313)
Prob >F .000 .009 Coalition 4.040
(2.749)
Note: p = .05, p = .001 (one-tailed test).
Coal Economy .807
(.347)
Electoral Rules 1.023
(2.277)
baseline results for concurrent and nonconcurrent elec- Rules Economy .856
tions. (.418)
The results again support my argument. In concur- Constant 5.130 8.657 5.973
rent legislative elections both the economy and the (4.561) (3.842) (3.611)
partys previous vote are significantly correlated with R2 .393 .381 .281
the vote in the current election. In contrast, Previous N 61 61 61
Vote matters less and Economy is not significant in Prob >F .000 .000 .006
nonconcurrent elections, although it is in the predicted Note: p = .05, p = .01, p = .001 (one-tailed test).
direction.16 This implies (but does not confirm, since I
have no variable testing for Local Conditions in both
models) that local conditions matter relatively more in
nonconcurrent elections and that the holding of both about half of what it would gain if the presidents party
executive and legislative elections simultaneously em- controlled a majority. This supports the clarity of re-
phasizes national policy matters in voters minds. sponsibility hypothesis that voters reward or punish
As with executive elections, the next question is the presidents party relatively less when that party
whether these results change when we add clarity of does not control a legislative majority.17
responsibility variables. I explore concurrent elections Model 2 tests for any effect in legislative elections
first, in Table 5. when the president governs in coalition (Coalition),
In Model 1, I test for any change when the presidents regardless of whether the coalition is majority or mi-
party does not control a majority of the seats (Minority nority. This situation arose in 18 elections. Similar
President). This hypothesis is the same as in Table 3. to Model 1, this hypothesis suggests that shared re-
This situation arose in 42 cases, and the null cases are sponsibility for government output dampens voters
thus those in which the presidents party controls more propensity to hold the incumbent presidents party
than 50% of the seats. The hypothesis suggests that responsible for the state of the economy. Thus, the
shared responsibility for government output dampens coefficient on the interacted variable again ought to
voters propensity to hold the incumbent presidents be negative, offsetting the positive expected coefficient
party responsible for the state of the economy in leg- on Economy. The result confirms the hypothesis. When
islative elections. Thus, the coefficient on the interacted the president governs in coalition, the economy still
variable ought to be negative, offsetting the expected affects the vote, but if the economy increased by 1%
positive coefficient on Economy. and the president governed in coalition, the presidents
The results indicate that the Previous Vote, Econ- party would only gain 0.952 0.807 = 0.147%, even
omy, and Reelection variables remain strongly corre- less than if the president lacks a majority legislative
lated with the vote and that the interaction of Minority contingent.18
President with the Economy variable is in the predicted
direction and is significant. To illustrate the finding,
17
suppose that the economy increased by 1%. If the This model excludes the 1995 election in Argentina as an outlier.
presidents party did not control a majority, then the Exclusion of no other case affects the results. When this case is in-
cluded, the Previous Vote and Economy variables remain significant,
presidents party would gain 1.211 0.562 = 0.649%, while the Reelection variable is not significant, as is the Minority
President Economy variable, although it is still in the predicted
direction. Excluding no other case alters the results.
16 For concurrent elections, if we exclude the 1990 Dominican 18 This model excludes the 1999 Panamanian election as an outlier.
Republic election, the Reelection variable becomes statistically sig- When this case is included, the Previous Vote and Economy variables
nificant. No other case affects the results. For the nonconcurrent remain significant, while the Reelection variable is not significant, as
elections, if we exclude the 1994 Ecuador election, the coefficient on is the Coalition Economy variable. The exclusion of one other case
Previous Vote becomes significant. No other case alters the results. affects the results, in a similar way: the 1998 Venezuelan election.

7
Presidentialism and Accountability for the Economy August 2004

Finally, Model 3 tests for the impact of the Electoral

Rules. Electoral systems that encourage localism or TABLE 6. Nonconcurrent Legislative
personalistic campaigning may tend to obscure the clar- Elections
ity of responsibility for the state of the national econ- Independent
Variable Model 1 Model 2 Model 3
omy, while nationalizing electoral systems might
Previous Vote .178 .193 .189
maintain the clarity of responsibility for the economy.
(.115) (.106) (.081)
I created a dummy variable designed to test whether Economy .260 .315 .603
the most extremely localizing and/or personalizing (.329) (.258) (.277)
electoral systems differ in a substantial way from other Minority President .096
electoral systems. These include the systems of Brazil, (2.386)
pre-1973 Chile, Colombia, Cyprus, Peru, Taiwan, and Min Pres Economy .185
Uruguay (Samuels and Shugart 2003).19 For concurrent (.439)
elections, only six cases took a value of one. As with Coalition 1.602
the other variables, we expect the coefficient on the (3.166)
interacted variable to be negative. Coal Economy .287
(.602)
The results again reveal the robustness of the first
Electoral Rules 7.722
two variables in the regression to the introduction of (3.724)
this variable. The interacted Rules Economy variable Rules Economy 1.314
is also significant, indicating that localizing electoral (.797)
rules diffuse clarity of responsibility for the economy. Constant 4.807 5.905 3.025
However, I should acknowledge that this result mer- (5.956) (5.784) (3.665)
its additional exploration, because this model excludes R2 .178 .182 .324
one of the six localizing cases, the 1990 Peruvian elec- N 41 41 41
tion, as an outlier. When this case is included, there is no Prob >F .019 .039 .022
relationship between this variable and the dependent Note: p = .05 (one-tailed test).
variable. However, the other variables in the regression
do not change. In short, although this test suggests that
localizing electoral rules affect clarity of responsibility
in concurrent elections, the result is inconclusive. elections, whereas local or regional factors may weigh
In sum, for concurrent legislative elections, I obtain more. This does not mean that national factors are (or
robust results that the economy is related to voters should be) irrelevant in nonconcurrent elections, it is
propensity to sanction incumbents and that the partys only a statement of their relative importance. Specif-
vote in the previous election is related to its vote in the ically, the hypothesis suggests that the coefficients on
current election. In addition, partisan variables such all of the variables in the regressions should be weakly
as coalition and minority government diffuse the re- significant or not significant. For example, the perfor-
lationship between the economy and the vote. These mance of the national party in the previous election
findings, combined with the findings in Tables 1 through should matter less if the current election is noncon-
3, suggest that in countries with a single-party majority current, weakening the coefficient on Previous Vote.
legislative contingent and concurrent elections, voters Likewise, the Economy ought to matter less or not at
reward or sanction executives and legislators for eco- all. Moreover, none of the interacted variables should
nomic performance similarly. However, party-system be significant, because we dont expect voters to take
fragmentation, minority government, and potentially national partisan factors into consideration in noncon-
the electoral rules can mediate this relationship and current elections.
obscure the clarity of responsibility for economic out- Let me first comment on the results for the Pre-
comes in legislative elections. vious Vote variable. Recall that in concurrent elec-
I now conduct similar tests for the nonconcurrent tions, this variable was always significant, and in the
legislative elections. The results appear in Table 6. Re- baseline model for nonconcurrent elections (Table 4),
call that I hypothesized that in general, the account- it remained significant, although the coefficient was
ability relationship between economics and elections much weaker than that for concurrent elections. Here,
should weaken in nonconcurrent legislative elections, the Previous Vote variable reaches significance in two
because national political factors matter relatively less of three models and its coefficient remains smaller
in nonconcurrent elections. All else equal, nonconcur- than for concurrent elections, a plausible finding given
rent elections focus voters and candidates attention our intuition that nonconcurrent elections tend to
on local factors and candidates personal qualities and concentrate campaigns and voters minds on local or
constituency service. Moreover, national party organi- candidate-based factors rather than national partisan
zations play less of a role in nonconcurrent legislative factors.
Next I explore the impact of the Economy variable.
In concurrent elections this variable was robust to all
19 My operationalization collapses Samuels and Shugarts (2003) tri-
alternative specifications. Here, Economy is significant
partite coding, because I found that no difference exists between the
cases they coded with a value of one and those they coded with two
in only one model, and it is always smaller in non-
but that a difference exists between the cases they coded as zero and concurrent elections than in any of the models for
all the other cases. concurrent elections. These results thus again conform

8
American Political Science Review Vol. 98, No. 3

to the expectations generated by Table 4, which con- presidentialism. The results confirm neither the most
trasted concurrent and nonconcurrent legislative elec- optimistic views, expressed most clearly (if only at
tions and found that the relationship between the an abstract level) by Persson, Roland, and Tabellini
economy and the vote was significant in both types (1997), nor the most pessimistic, perhaps expressed
but both weaker and less robust in nonconcurrent by Linz (1990, 1994). In many ways the findings echo
elections. Shugart and Careys (1992) suggestion that the sep-
Finally, I examine the impact of the other clarity of aration of powers can engender different forms of
responsibility variables in Table 6. Recall that I argued representation and accountability than parliamentary
that nonconcurrence tends to reduce the impact of systems.
national party organization and campaign messages. Clarity of responsibility and thus voters capacity to
Thus the influence of all national-level partisan factors sanction elected officials work in particular ways under
ought to be relatively diminished in nonconcurrent presidentialism. When elections are nonconcurrent,
elections. The results are clear: In contrast to the re- a situation that cannot occur under parliamentarism,
sults in Table 5, none of the interacted variables are sanctioning for the state of the economy is weak. In ex-
significant in Table 6, indicating that no other clarity ecutive elections this lack of a connection between the
of responsibility variables mediate the impact of the economy and the vote came somewhat as a surprise, but
national economy beyond the electoral cycle.20 the relatively weak relationship in legislative elections
To sum up, the results for legislative elections in followed theoretical expectations and research findings
Tables 46 indicate that the economy matters in on U.S. elections. Because about one in five executive
both concurrent and nonconcurrent legislative elec- elections and two in five legislative elections around the
tions, although it has a greater impact and matters world are nonconcurrent, these results support some
more clearly and consistently in concurrent elections. observers concerns about overall accountability under
Moreover, the results indicate that partisan and in- presidentialism.
stitutional clarity of responsibility affects the rela- In contrast, concurrence enhances voters capacity
tionship between the economy and the vote in con- to reward or sanction government officials for the state
current elections but that the evidence for such a of the economy in presidential systems. These results
connection is much weaker in nonconcurrent leg- qualify Linzs critique of presidentialism based on his
islative elections. In addition, when we compare the notion of dual democratic legitimacies. Despite the
impact of the economy in executive and legisla- formal separation of powers, institutions that promote
tive elections (compare against Tables 1 and 3, e.g.) close electoral linkage between the executive and
we can infer that voters reward or sanction presidents legislative branches can generate unified democratic
more than legislators for economic performance.21 legitimacy. Ceteris paribus, when elections are con-
These results leave us with the following: In concur- current voters treat the incumbent executive and his or
rent executive elections, the economy affects the vote, her legislative supporters as a team and judge them as
regardless of the clarity of responsibility. Yet in noncon- such.
current executive elections, the economy never mat- However, the all-else-equal clause is essential, be-
ters. In concurrent legislative elections, the economy cause in concurrent elections legislative fragmentation
affects the vote, but partisan and institutional variables and localizing electoral rules can affect the degree
mediate that connection. The economy also affects the to which voters hold both branches accountable for
vote in nonconcurrent legislative elections, but rela- the economy. This implies the following: Under single-
tively less so than in concurrent legislative elections, party majority government with concurrent elections,
yet neither partisan nor institutional variables mediate voters in presidential and parliamentary systems will
that link. These findings tend to support my argument not differ much in their capacity to sanction or reward
about the particular way in which the electoral cycle governments. However, even under concurrent elec-
affects clarity of responsibility and thus accountability tions, when we see multiparty coalitions (for exam-
in presidential systems. ple), clarity of responsibility and thus voters ability to
sanction incumbents function differently under pres-
identialism and parliamentarism. Under parliamen-
CONCLUSION tarism, as Anderson (1995, 2000), Powell and Whitten
This paper elucidates the conditions under which elec- (1993), and others have argued, partisan and institu-
toral sanctioning is more or less likely to occur under tional variables tend to diffuse clarity of responsibility.
Under presidentialism, the same phenomenon occurs
in concurrent legislative elections but not in executive
20 The negative and significant result on Electoral Rules indicates
elections. This separates the degree to which voters
that candidates from the presidents party obtain a higher percentage
of the vote when the electoral rules localize competition. This
sanction executives and legislators and means that
occurs because localizing electoral rules allows candidates to insulate executives in many presidential systems are subject to
themselves from national trends and base their campaigns on the relatively greater sanctioning or rewards than legisla-
personal vote and the incumbency advantage. tors, a situation that cannot by definition arise under
21 Regression analysis that pooled all elections, however, did not
parliamentarism.
statistically distinguish between the effect of the economy in execu-
tive and that in legislative elections, although the sign of the dummy
Presidentialism generates distinct forms of account-
variable distinguishing executive from legislative elections was in the ability. The electoral cycle most clearly affects this dy-
predicted direction. namic, but electoral rules and party system features

9
Presidentialism and Accountability for the Economy August 2004

also mediate voters ability to sanction incumbents. in ways that parliamentary elections cannot. Thus al-
Under a limited set of conditions presidentialism can though the state of the national economy matters rela-
approximate parliamentarism, and we would expect tively little in nonconcurrent legislative elections, and
to see a similar form of government accountability national clarity of responsibility variables do not af-
in both types of systems. But the collective, respon- fect this relationship, local variables may systemati-
sible parties, policy-oriented sense of accountability cally affect the vote and thus promote accountability
that is rooted in normative political theory can oc- (for example, whether voters can recall the incum-
cur under presidentialism only under these limited bents name or the extent to which they credit
conditions. When such conditions are not met, and the incumbent with providing constituency service).
when the electoral and party systems separate the The degree to which presidentialism allows both na-
bases for selecting members of each branch, we have tional and local accountability simultaneously merits
a more clearly distinct presidentialist dynamic. For exploration.
example, even when elections are concurrent, vot- Yet if one disputes Shugart and Careys suggestion
ers are often likely to sanction incumbents in both that separating the bases for electing the executive and
branches of government to different degrees, depend- legislative branches can be a normatively good thing
ing on institutional and party-system factors. And for democracy because it permits different forms of ac-
when elections are nonconcurrent, national factors countability, and instead holds to the Linzian view that
matter relatively little in both executive and legislative both branches should generally be held accountable
elections. for the same things, then the results here reinforce the
For legislative elections at least, Shugart and Carey conclusion that the core problem of for government ac-
(1992) suggested that nonconcurrence and/or localiz- countability under presidentialism is the potential lack
ing electoral rules can be a normatively good thing of electoral linkage between the president and his or
because there is nothing prima facie wrong with the her legislative contingent. Party-system and other insti-
idea of giving legislators an opportunity to campaign tutional factors shape the strength of this linkage, so
on their constituency service and voters an opportu- the findings in this paper thus reinforce the critical
nity to base their vote on local rather than national role that institutions and party-system configurations
concerns. In this way the institutional separation of can play in shaping democratic performance. Future
purpose under presidentialism (Samuels and Shugart research ought to explore these differences in more
2003) can enhance representation and accountability detail.

APPENDIX 1: CASES

Country Electionsa
Argentina 1983PL, 1985L, 1987L, 1989PL, 1991L, 1993L, 1995PL, 1997L, 1999PL
Benin 1995L, 1999L
Bolivia 1985PL, 1989PL, 1993PL, 1997PL
Brazil 1989P, 1990L, 1994PL, 1998PL
Chile 1958P, 1961L, 1964P, 1965L, 1969L, 1970P, 1973L, 1989PL, 1993PL, 1997L, 1999P
Colombia 1974PL, 1978PL, 1982PL, 1986PL, 1990PL, 1991L 1994PL, 1998PL
Costa Rica 1962PL, 1966PL, 1970PL, 1974PL, 1978PL, 1982PL, 1986PL, 1990PL, 1994PL, 1998PL
Cyprus 1981L, 1985L, 1991L, 1996L, 2001L
Dominican Republic 1986PL, 1990PL, 1994PL, 1996P, 1998L
Ecuador 1979L, 1984PL, 1986L, 1988PL, 1990L, 1992PL, 1994L, 1996PL, 1998L, 2000P
El Salvador 1989PL, 1991L, 1994PL, 1997L, 1999P, 2000L
Guatemala 1991PL, 1995PL, 1999PL
Honduras 1981PL, 1985PL, 1989PL, 1993PL, 1997PL
Korea 1987P, 1988L, 1992PL, 1996L, 1997P, 2000L
Malawi 1994PL, 1999PL
Mexico 1994PL, 1997L, 2000PL
Nicaragua 1991PL, 1996PL
Panama 1994PL, 1999PL
Paraguay 1989PL, 1993PL, 1998PL
Peru 1985PL, 1990PL
United States 1962L, 1964PL, 1966L, 1968PL, 1970L, 1972PL, 1974L, 1976PL, 1978L, 1980PL, 1982L,
1984PL, 1986L, 1988PL, 1990L, 1992PL, 1994L, 1996PL, 1998L, 2000PL
Uruguay 1966PL, 1971PL, 1984PL, 1989PL, 1994PL, 1999PL
Venezuela 1963PL, 1968PL, 1973PL, 1978PL, 1983PL, 1988PL, 1993PL, 1998PL, 2000PL
Note: More cases are listed here than are included in the regressions. None of the first elections in any time series are included because
there is no t 1 information on the Previous Vote variable for the first. Thus the total number of cases included in the regressions equals
the total number of cases listed here minus the first cases in every countrys series.
a P, presidential; L, legislative.

10
American Political Science Review Vol. 98, No. 3

APPENDIX 2: ELECTORAL DATA SOURCES REFERENCES

For Argentina, Bolivia, Brazil, Chile, Costa Rica, the Domini- Anderson, Christopher. 1995. The Dynamics of Public Support
can Republic, Ecuador, El Salvador, Guatemala, Honduras, for Coalition Governments. Comparative Political Studies 28
Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and (Summer): 35083.
Venezuela, the sources were as follows. Anderson, Christopher. 2000. Economic Voting and Political Con-
text: A Comparative Perspective. Electoral Studies 19 (June):
r Inter-American Development Bank. 2002. Democracies 15170.
in Development. Washington, DC: IDB. Carey, John M., and Matthew S. Shugart. 1995. Incentives to Cul-
r Political Database of the Americas. 2003. Statisti- tivate a Personal Vote: A Rank-Ordering of Electoral Formu-
las.Electoral Studies 14 (December): 41739.
cal Data: Electoral Results. Various dates. http://www. Cheibub, Jose Antonio, and Adam Przeworski. 1999. Democ-
georgetown.edu/pdba/. Retrieved November 2003.
r Magar, Eric, and Kevin J. Middlebrook. 2000. Statisti-
racy, Elections, and Accountability for Economic Outcomes.
In Democracy, Accountability, and Representation, ed. Adam
cal Appendix: National Election Results, 1980s and 1990s, Przeworski, Susan Stokes, and Bernard Manin. New York:
for Argentina, Brazil, Chile, Colombia, El Salvador, Peru, Cambridge University Press, 22250.
and Venezuela. In Conservative Parties, the Right, and Dorussen, Han, and Harvey Palmer. 2002. The Context of Eco-
Democracy in Latin America, ed. Kevin Middlebrook. nomic Voting: An Introduction. In Economic Voting, ed. Han
Baltimore: Johns Hopkins University Press. Dorussen and Michaell Taylor. London and New York: Routledge,
114.
For particular countries, I also used the following. Elgie, Robert. 1999. The Politics of Semi-presidentialism. In Semi-
Presidentialism in Europe, ed. Robert Elgie. Oxford: Oxford Uni-
r Benin: personal correspondence, Professor Shaheen versity Press, 121.
Mozaffar, Bridgewater State University. Erikson, Robert S. 1989. Economic Conditions and the Presidential
r Bolivia: Gamarra, Eduardo, and James Malloy. 1995. The Vote. American Political Science Review 83 (June): 56773.
Erikson, Robert S. 1990. Economic Conditions and the Congres-
Patrimonial Dynamics of Party Politics in Bolivia. In sional Vote: A Review of the Macrolevel Evidence. American
Building Democratic Institutions: Party Systems in Latin Journal of Political Science 34 (May): 37399.
America, ed. Scott Mainwaring and Timothy R. Scully, Fiorina, Morris. 1981. Retrospective Voting in American National
Stanford, CA: Stanford University Press.
r Brazil: Jairo Nicolau. 2003. Dados Eleitorais do Brasil.
Elections. New Haven, CT: Yale University Press.
Inter-American Development Bank. 2002. Democracies in Develop-
www.iuperj.br/deb/port/. Retrieved November 2003. ment. Washington, DC: IDB.
r Chile: (a) Chilean government Web site, www.elecciones. Lewis-Beck, Michael. 1988. Economics and Elections: The Major
gov.cl. Retrieved November 2003. (b) Cruz-Coke, Western Democracies. Ann Arbor: University of Michigan Press.
Bernardo. 1984. Historia Electoral de Chile, 19251973. Lewis-Beck, Michael. 1997. Whos the Chef? Economic Voting
Santiago: Editorial Judicial de Chile. Under a Dual Executive. European Journal of Political Research
r Costa Rica: personal correspondence, Michelle Taylor-
31 (April): 31525.
Lewis-Beck, Michael, and Mary Stegmeier. 2000. Economic De-
Robinson, Texas A&M University.
r Cyprus: Nordsieck, Wolfram. 2003. Parties and Elec-
terminants of Electoral Outcomes. Annual Review of Political
Science 3: 183219.
tions in Europe. www.parties-and-elections.de/indexe. Lijphart, Arend. 1992. Introduction. In Parliamentary versus Pres-
html. Retrieved November 2003. idential Government, ed. Arend Lijphart. Oxford: Oxford Univer-
r Ecuador: Meja-Acosta, Andres. 2000. Weak Coali- sity Press, 130.
tions and Policy Making in the Ecuadorian Congress Linz, Juan. 1990. The Perils of Presidentialism. Journal of Democ-
(19791996). Presented at the 2000 meeting of the Latin racy 1 (Winter): 5169.
American Studies Association, Chicago. Linz, Juan J. 1994. Presidential versus Parliamentary Democracy:
r Korea: (a) Keesings Contemporary Archives (various). Does It Make a Difference? In The Failure of Presidential Democ-
racy, Volume 2: The Case of Latin America, ed. Juan Linz and
(b) Foreign Broadcast Information Service. 1992 Daily Arturo Valenzuela. Baltimore: Johns Hopkins University Press,
Report, East Asia. March 26, 1992, pp. 2223. (c) Moriss, 390.
Peter. 1996. Electoral Politics in South Korea. Electoral Manin, Bernard, Adam Przeworski, and Susan Stokes. 1999. Elec-
Studies 15 (December): 55062. (d) Kang, W. T., and H. tions and Representation. In Democracy, Accountability, and
Jaung. 1999. The 1997 Election in Korea. Electoral Stud- Representation, ed. Adam Przeworski, Susan C. Stokes, and
ies 18 (December): 599608. Bernard Manin. New York: Cambridge University Press, 2954.
r Malawi: (a) Elections Around the World. 2003. Nannestad, Peter, and Martin Paldam. 2002. The Cost of Ruling. In
www.electionworld.org. Retrieved November 2003. Economic Voting, ed. Han Dorussen and Michaell Taylor. London
(b) Nohlen, Dieter, Michael Krennerich, and Bernhard and New York: Routledge, 1744.
ODonnell, Guillermo. 1994. Delegative Democracy. Journal of
Thibaut, ed. 1999. Elections in Africa: A Data Handbook.
Democracy 5 (January): 5569.
Oxford: Oxford University Press.
r Panama: personal correspondence, Carlos Guevara-
Pacek, Alexander, and Benjamin Radcliff. 1995. The Political Econ-
omy of Competitive Elections in the Developing World. Ameri-
Mann, University of Notre Dame.
r Peru: Tuesta Soldevilla, Fernando. 2001. Peru Poltico en
can Journal of Political Science 39 (August): 74559.
Palmer, Harvey D., and Guy D. Whitten. 1999. The Electoral Impact
Cifras: 18212001. Lima: Fundacion Friedrich Ebert. of Unexpected Inflation and Economic Growth. British Journal
r Uruguay: No author. 2000. Elecciones 1999/2000. of Political Science 29 (October): 62339.
Montevideo: Ediciones de la Banda Oriental. Persson, Torsten, Gerard Roland, and Guido Tabellini. 1997. Sepa-
r United States: Ornstein, Norman, et al. Vital Statistics on ration of Powers and Political Accountability. Quarterly Journal
Congress, 20012002. Washington, DC: American Enter- of Economics 112 (November): 1163203.
Powell, G. Bingham, and Guy D. Whitten. 1993. A Cross-National
prise Institute.
r Venezuela: Kornblith, Miriam, and Daniel H. Levine.
Analysis of Economic Voting: Taking Account of the Political
Context. American Journal of Political Science 37 (May): 391
1995. Venezuela: The Life and Times of the Party Sys- 414.
tem. In Building Democratic Institutions: Party Systems Remmer, Karen L. 1991. The Political Impact of Economic Crisis
in Latin America, ed. Scott Mainwaring and Timothy R. in Latin America in the 1980s. American Political Science Review
Scully. Stanford, CA: Stanford University Press. 85 (September): 777800.

11
Presidentialism and Accountability for the Economy August 2004

Remmer, Karen L. 2001. Elections and Economics in Contem- Debate. In Presidentialism and Democracy in Latin America, ed.
porary Latin America. Presented at the meeting of the Latin Scott Mainwaring and Matthew Shugart. New York: Cambridge
American Studies Association, Chicago. University Press, 1254.
Samuels, David, and Matthew Shugart. 2003. Presidentialism, Siaroff, Alan. 2003. Comparative Presidencies: The Inadequacy
Elections, and Representation. Journal of Theoretical Politics of the Presidential, Semi-Presidential and Parliamentary Dis-
15 (January): 3360. tincion. European Journal of Political Research 42: 287
Shugart, Matthew S. 1995. The Electoral Cycle and Institutional 312.
Sources of Divided Presidential Government. American Political Stokes, Susan. 2001. Introduction. In Public Support for Market
Science Review 89 (June): 32743. Reforms in New Democracies, ed. Susan C. Stokes, New York:
Shugart, Matthew S., and John M. Carey. 1992. Presidents and Assem- Cambridge University Press, 134.
blies: Constitutional Design and Electoral Dynamics. New York: Strm, Kaare. 1990. Minority Government and Majority Rule. New
Cambridge. York: Cambridge.
Shugart, Matthew S., and Scott Mainwaring. 1997. Presidentialism World Bank. 2003. World Development Indicators [CD-ROM].
and Democracy in Latin America: Rethinking the Terms of the Washington, DC: World Bank.

12